[RFC] Case-preserving replacement sequence

[Hu Jialun]

Hello,

I would like to implement a new feature as detailed below, namely a new 
backslash sequence in replacement for case-preserving replacement, and 
would like to hear some comments before proceeding.

The sequence, which shall be denoted as \P tentatively, would be 
followed by a capture group number, e.g. \P4. This would normalise the 
capitalisation of following n characters (where n is the length of the 
corresponding capture group) to that of the capture group. Such a 
feature would supposedly offer a easier and faster way to do 
case-preserving replacements than, say, a Perl script with replacement 
callback.

As a trivial example,

        $ sed 's/\(REGular\)/\P1expression/i' <<< "regular expression"
        EXPression expression

        $ sed 's/\(his\)\([[:punct:]]\)*\(word\)/\P1her\2\P3name/i' <<-EOF
                his,word
                His.WORD
                his-wOrD
                EOF
        her,name
        Her.NAME
        her-nAmE

Any feedback would be greatly appreciated.

Many thanks,
Hu Jialun